Histogram-based radiomic texture analysis on high-resolution computed tomography (HRCT) images can accurately differentiate middle ear cholesteatoma from non-cholesteatomatous lesions, according to a retrospective study published in BMC Medical Imaging. The approach achieved strong diagnostic accuracy and high inter-observer reproducibility when evaluating temporal bone scans.

Researchers Eren Ozgur and Murat Ozogul evaluated the diagnostic accuracy of texture analysis on temporal bone HRCT scans to determine whether quantitative radiomic features could serve as a complementary diagnostic tool. According to the authors, this quantitative imaging strategy could assist clinicians during preoperative assessments, particularly in patients presenting with diagnostic uncertainty or those with contraindications to magnetic resonance imaging (MRI).

The retrospective investigation analyzed 96 age- and sex-matched patients who underwent temporal bone HRCT examinations between 2020 and 2026. The study cohort comprised 48 individuals with surgically or histopathologically confirmed cholesteatoma and 48 patients diagnosed with non-cholesteatomatous middle ear pathologies. Three-dimensional volumes of interest were manually segmented on axial computed tomography sections, and histogram-based radiomic features were subsequently extracted using LIFEx software.

To determine the most effective diagnostic parameters, the investigators constructed a multivariable binary logistic regression model using a backward elimination method. Four independent histogram features were identified and incorporated into the predictive model: Intensity Histogram Mean Absolute Deviation and Minimum Grey Level, which were both significantly lower in the cholesteatoma group, alongside Intensity Histogram Kurtosis and Range, which were both significantly higher among patients with cholesteatoma.

In receiver operating characteristic curve analysis, the predictive model demonstrated favorable calibration with a Hosmer–Lemeshow p-value of 0.947 and achieved an area under the curve (AUC) of 0.912. Internal validation conducted through five-fold cross-validation, with variable selection repeated independently within each training fold, maintained an AUC of 0.905 and an overall diagnostic accuracy of 82.0 percent. Inter-observer reliability assessment yielded an intraclass correlation coefficient of 0.905, confirming consistent reproducibility across independent segmentations.

Subject of Research: Radiomic histogram analysis of temporal bone HRCT for cholesteatoma diagnosis

Article Title: A quantitative radiomic approach to cholesteatoma using histogram analysis

Article References: Ozgur, E., & Ozogul, M. (2026). A quantitative radiomic approach to cholesteatoma using histogram analysis. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-03033-4

Image Credits: AI Generated

DOI: 10.1186/s12880-026-03033-4

Keywords: cholesteatoma, radiomics, computed tomography, histogram analysis, middle ear, texture analysis, medical imaging, HRCT, quantitative, radiomic, approach, histogram

News Source: Ophelia Keating. (November 6, 2026). CT Histogram Radiomics Accurately Distinguishes Middle Ear Cholesteatoma. Scienmag.